Reported values for the heat of activation,H, for volume chemical diffusion, as obtained from the slope of the best straight line in a semilogarithmic plot of diffusion data againstT−1, may often be greatly in error. These values will be too low whenever diffusion along internal surfaces plays a significant role at the lower temperatures of measurement. A method is presented which attempts to avoid such errors inHby using the calculated values of Zener for the intercept of the straight line. When the data is reconsidered in the light of this method, the widely accepted belief thatHfor chemical diffusion in dilute solid solutions is generally considerably less thanHfor self‐diffusion of the solvent is brought to question. In fact, it is found that the chemical diffusion activation energy is generally within 15 percent of that for self‐diffusion. It is further concluded that the process of diffusion under a chemical gradient may itself produce internal surfaces which act as short‐circuiting diffusion paths.